{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/chauffeurnet-learning-to-drive-by-imitating","title":"ChauffeurNet: Learning to Drive by Imitating the Best and Synthesizing the Worst","arxiv_id":"1812.03079","date":"2018-12-07","proceeding":null,"authors":["Mayank Bansal","Alex Krizhevsky","Abhijit Ogale"],"abstract":"Our goal is to train a policy for autonomous driving via imitation learning\nthat is robust enough to drive a real vehicle. We find that standard behavior\ncloning is insufficient for handling complex driving scenarios, even when we\nleverage a perception system for preprocessing the input and a controller for\nexecuting the output on the car: 30 million examples are still not enough. We\npropose exposing the learner to synthesized data in the form of perturbations\nto the expert's driving, which creates interesting situations such as\ncollisions and/or going off the road. Rather than purely imitating all data, we\naugment the imitation loss with additional losses that penalize undesirable\nevents and encourage progress -- the perturbations then provide an important\nsignal for these losses and lead to robustness of the learned model. We show\nthat the ChauffeurNet model can handle complex situations in simulation, and\npresent ablation experiments that emphasize the importance of each of our\nproposed changes and show that the model is responding to the appropriate\ncausal factors. Finally, we demonstrate the model driving a car in the real\nworld.","url_abs":"http://arxiv.org/abs/1812.03079v1","url_pdf":"http://arxiv.org/pdf/1812.03079v1.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"chauffeurnet-learning-to-drive-by-imitating","repo_url":"https://github.com/SSN15/Behavioral-Cloning--Implementaion-of-Autonomous-car-using-deep-learning","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"chauffeurnet-learning-to-drive-by-imitating","repo_url":"https://github.com/SSN15/Traffic-sign-recognition-using-deep-learning-and-computer-vision","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"none","reach":{"status":"ok"}},{"paper_slug":"chauffeurnet-learning-to-drive-by-imitating","repo_url":"https://github.com/aidriver/ChauffeurNet","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok"}},{"paper_slug":"chauffeurnet-learning-to-drive-by-imitating","repo_url":"https://github.com/shunchan0677/deepware","is_official":0,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"tf","reach":{"status":"ok","spdx":"MIT"}}],"tasks":[{"task_slug":"autonomous-driving","task_name":"Autonomous Driving"},{"task_slug":"imitation-learning","task_name":"Imitation Learning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[],"syntology":{"syntology_url":"https://syntology.ai/paper/1812.03079","atlas_url":"https://app.syntology.ai/?focus=1812.03079","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}